Private Schools in the United States: A Statistical Profile, 1993-94 / Appendix B: Technical Notes

Appendix B:  Technical Notes

Survey Content

The Schools and Staffing Survey (SASS) consists of four main component sur veys administered to public school districts, and schools, principals, and teach ers in public and private schools. These surveys are the Teacher Demand and Shortage Survey, the School Principal Survey, the School Survey, and the Teacher Survey.

Copies of the questionnaires used in the SASS can be obtained by writing to:

Target Population and Estimates for SASS

A. Target Populations

The target populations for 199394 SASS were:

B. Estimates

The SASS was designed to support estimates at both the state and national level for the public sector, and at the national and association level for the private sector. The association groups for private schools were determined by the schools association or affiliation group listed on the 199192 Private Schools Survey (the frame) and updated with 199293 association lists. The association groups, i.e., the private school types reported in the top part of the tables in the Tables Section and in the graphic profiles by affiliation category, were determined in the following order:

1) Militarymembership in the Association of American Military Colleges and Schools;

2) Catholicaffiliation as Catholic or membership in the National Catholic Education Association or the Jesuit Secondary Education Association;

3) Friendsaffiliation as Friends or membership in the Friends Council on Education;

4) Episcopalaffiliation as Episcopal or membership in the National Association of Episcopal Schools;

5) Hebrew Daymembership in the National Society for Hebrew Day Schools;

6) Solomon Schechtermembership in the Solomon Schechter Day Schools;

7) Other Jewishother Jewish affiliation; 8) Missouri Synodmembership in the Lutheran Church, Missouri Synod;

9) Wisconsin Synodmembership in the Evangelical Lutheran Church - Wisconsin Synod or affiliation as Evangelical Lutheran - Wisconsin Synod;

10) Evangelical Lutheranmembership in the Association of Evangelical Lutheran Churches or affiliation as Evangelical Lutheran Church in America;

11) Other Lutheranother Lutheran affiliation;

12) Seventh-Day Adventistaffiliation as Seventh-Day Adventist or membership in the General Conference of Seventh-Day Adventists;

13) Christian Schools Internationalmembership in Christian Schools International;

14) Association of Christian Schools Internationalmembership in the Association of Christian Schools International;

15) National Association of Private Schools for Exceptional Children membership in the National Association of Private Schools for Exceptional Children;

16) Montessorimembership in the American Montessori Society or other Montessori associations;

17) National Association of Independent Schoolsmember of the National Association of Independent Schools;

18) National Independent Private School Associationmember of the National Independent Private School Association;

19) All Other Private Schoolsmember of any other association specified in the PSS or affiliated with a group not listed above or not a member of any association.

Schools with multiple affiliations were classified by their first affiliation in the list above. For consistency with the typology, one exception was made. One school with a religious affiliation other than Catholic that reported membership in the National Catholic Education Association was not counted as a Catholic school. One affiliation category, NIPSA, was newly added in 1993- 94. Although it will be included as a separate category in future SASS private school reports, NIPSA schools were grouped in the "all other" private schools in this report.

The grade levels of schools are defined in the following way:

As used in this report, middle schools refer to those elementary, secondary, and combined schools whose lowest grade is 4 or higher and whose highest grade is 7, 8, or 9.

Comparisons between public and private schools are only possible at the national and regional levels, because private schools are selected for sampling by association group and not by geographic location, such as state.

Sample Design and Implementation

A. Sampling Frame for Private Schools

The sampling frame for private schools was the 199192 Private School Survey, updated with 199293 association lists (Broughman et al. 1994). This data collection uses two components to develop estimates of the number of private schools in the United States. A list frame was the primary private school frame and an area frame was used to identify schools not included on the list frame and thereby compensate for the undercoverage of the list frame. (See Abramson et al. 1996 for a detailed description of the sample design.)

B. Sample Selection Procedures

Schools are the primary sampling unit in SASS. Public schools were selected to be representative at the national and state levels; private schools were selected to be representative at the national and association levels. More detail is available in Abramson et al. (1996).

Each selected school was asked to provide a list of their teachers and selected characteristics. Nine percent of the private schools and four percent of the public schools did not provide teacher lists. A factor in the teacher weighting system was used to adjust for these nonparticipant schools.

C. Sample Sizes for Private Schools

Table III.1 shows the sample size and number of interview cases for each questionnaire, by private school typology.

The number in sample is the number of in-scope, or eligible cases. This number excludes the out-of-scope cases, which are drawn for the sample but are not eligible for interview. For example, a school which has closed or a teacher who has left the country would be considered out-of-scope.

The number of interviews is the number of in-scope (eligible) cases minus the noninterview cases. The noninterview cases include refusals or sample questionnaires with too little valid data to be considered complete interviews for the survey. The number of interviews is the actual unweighted number of cases upon which estimates in this report are based. A nonresponse adjustment is included in the weights to reduce the bias due to nonresponse.

Table III.1 Number of in-scope sample cases and number of interviews, private school districts, principals, and schools: SASS 199394

                                    Private schools            Private principals             Private teachers
                               -------------------------    -------------------------     -------------------------
                               # in sample     # intvws.    # in sample     # intvws.     # in sample     # intvws.
All private schools               3,074          2,585          3,143          2,722          10,386         8,372
Catholic                            921            818          1,023            831           3,680         3,061
 Parochial                          465            408            462            427           1,776         1,474
 Diocesan                           290            263            290            244           1,192           988
 Private Order                      166            147            271            160             712           599
Other Religious                   1,419          1,151          1,394          1,236           4,404         3,483
 Conservative Christian             325            248            322            274             929           667
 Affiliated                         708            574            702            631           2,239         1,790
 Unaffiliated                       386            329            370            331           1,236         1,026
Nonsectarian                        734            616            726            655           2,302         1,828
 Regular program                    366            297            364            321           1,279          1036
 Special emphasis                   182            150            176            160             582           436
 Special education                  186            169            186            174             441           356
NOTE: The number of in-scope cases in sample excludes out-of-scope, or ineligible, cases. Reasons for a
school, principal, or teacher to be out-of-scope include school closure, principal or teacher leaving the school.
SOURCE: U.S. Department of Education, , Schools and Staffing Survey,
199394(School Questionnaire, Principal Questionnaire, and Teacher Questionnaire).There are still other reasons for
a case to be considered out-of-scope.

Data Collection Prodedures

Data collection operations for the 199394 SASS took place during the 199394 school year. Table IV.1 depicts both the specific data collection activity and the time frame in which it occurred.

Table IV.1 Data collection time schedule

                    Activity                                   Date of activity
Introductory letters mailed to school districts                September 1993
Introductory letters and teacher listing sheets mailed         October 1993
to schools
Census field representatives called school districts to        October 1993
obtain the name of a contact person to whom the Teacher 
Demand and Shortage questionnaire should
be addressed
Lists of teachers provided by schools                          OctoberDecember 1993
First mailing of questionnaires to school districts and        December 1993
school principals
First mailing of questionnaires to schools and to              JanuaryFebruary 1994
Second mailing of questionnaires to districts and              January 1994
school principals
Second mailing of questionnaires to schools and                FebruaryMarch 1994
Telephone follow-up of mail nonrespondents                     MarchJune 1994

Response Rates

A. Private Sector Survey Response Rates

The weighted response rates for each component of SASS are detailed in Table V.1. Table V.1 lists private response rates by private school typology for administrators, schools, and teachers. The response rates are useful as an indication of possible nonresponse bias. The weighted response rates were derived by dividing the sum of the basic weights for the interview cases by the sum of the basic weights for the eligible cases. The basic weight for each sample case was assigned at the time of sampling and is the inverse of the probability of selection.

Table V.1 Final weighted response rates by private school type for private school administrators, schools, and teachers

Private school type                    Principals     Schools       Teachers
All private schools                       87.6%         83.2%         80.2%
Catholic                                  92.4          88.8          83.2
 Parochial                                92.4          88.0          83.2
 Diocesan                                 93.3          90.9          82.7
 Private order                            89.4          87.9          84.2
Other religious                           82.7          77.5          75.0
 Conservative Christian                   82.6          76.5          70.1
 Affiliated                               81.9          76.5          75.4
 Unaffiliated                             83.6          79.5          80.5
Nonsectarian                              89.7          86.1          81.6
 Regular                                  90.6          86.4          82.7
 Special emphasis                         89.0          81.4          78.0
 Special education                        88.5          93.2          81.1
SOURCE: U.S. Department of Education, National Center for Education
Statistics, Schools and Staffing Survey, 199394 (Principal Questionnaire,
School Questionnaire, and Teacher Question-naire).
Teacher response rates refer to the percentage of teachers responding in schools that provided teacher lists for sampling. Nine percent of private schools and 4 percent of public schools did not send in teacher lists. The effective response rate is calculated by multiplying together the teacher list rate and the response rate:

Private teachers: (.91 x .801) x 100 = .7289 x 100 = 72.9 percent effective response rate

Responding teachers were weighted to represent all teachers. Some teach ers responded even though no school questionaire was returned from their school. Responses from those teachers are included in the main graphic pro files. However, because the 18-category private school type was not deter mined for these teachers, their responses were not included in the Tables Section or in the graphic profiles by affiliation.

B. Item Response Rates

The unweighted item response rates (i.e., the number of sample units responding to an item divided by the number of sample units that participated in the survey) for the SASS and the Library Survey ranged from 50 percent to 100 percent. Tables V.2 and V.3 provide a brief summary of the item response rates. The item response rates in these tables are unweighted, and do not reflect additional response loss due to respondents refusal to participate in the survey.

Table V.2 Summary of unweighted private sector item response rates by questionnaire

                                          % of items         % of items
                          Range of      with a response    with a response
                        item response       rate of          rate of less
Survey                     rates          90% or more          than 75%
Principal Survey          55-100%             90%                  6%
School Survey             61-100%             77%                  3%
Teacher Survey            69-100%             89%                  1%
Student Survey            84-100%             97%                  0%
SOURCE: U.S. Department of Education, National Center for Education
Statistics, Schools and Staffing Survey, 1993-94 (Principal Questionnaire,
School Questionnaire, and Teacher Question-naire).
Table V.3 Items with private sector response rates of less than 75 percent*
    Survey                            Items
Principal Survey           14b(1,1), 14b(2,1), 14b(4,1),
                           14b(5,1), 14b(8,1), 21a, 21c, 28b
School Survey              31c(2), 31c(5), 31c(6), 31c(7),
                           31c(8), 31c(9)
Teacher Survey             39, 51c, 55
*/ The questionnaire wording for these items can be found in
The Schools and Staffing Surveys: 199394, Data File Users
Manual, a forthcoming NCES publication.
SOURCE: U.S. Department of Education, National Center for
Education Statistics, Schools and Staffing Survey, 199394
(Principal Questionnaire, School Questionnaire, and Teacher
Item 14b concerns principals previously held school positions. Items 21a and c concern principals eligibility and plans to retire. Item 28b concerns the tribal enrollment of American Indian or Alaska Native principals. Item 31c concerns the difficulty of filling vacancies in various fields. Item 39 concerns teachers number of classes/sections taught per week. Item 51c concerns teachers plans to retire from teaching. Item 55 concerns teachers total comp-bined family income.

Imputation Procedures

For questionnaire items that should have been answered but were not, values were imputed by (1) using data from other items on the questionnaire, (2) extracting data from a related component of the Schools and Staffing Survey (for example, using data from a school record to impute missing values on that schools LEA questionnaire), (3) extracting data from the sample file (information about the sample case from other sources; for example, the Private School Survey or the Common Core of Data, collected in the 199192 school year), and (4) extracting data from a respondent with similar character-istics.

For some incomplete items, the entry from another part of the questionnaire or information from the sample file was directly imputed to complete the item; for others the entry was used as part of an adjustment factor with other data on the incomplete record. For example, if a respondent did not report whether a school offered remedial reading in item 22a of the public school questionnaire, the response (1 = Yes or 2 = No) for a similar school was imputed to item 22a of the incomplete record. However, if a respondent had answered "Yes" to item 22a but had not reported the number of students in the program, the ratio of number of students in remedial reading to the total enrollment for a similar school was used with the enrollment at the school for which item 22a was incomplete to impute an entry to item 22a (i.e., SCHOOL A item 22a = SCHOOL A ENROLLMENT multiplied by the ratio of SCHOOL B item 22a to SCHOOL B ENROLLMENT).

Values were imputed to items with missing data for records that had been classified as interviews (ISR=1). Noninterview adjustment factors were used during the data weighting process to compensate for data that were missing because the sample case was a noninterview (ISR=2). For more information about imputation procedures, see Abramson et al. (1996).


The private sector was weighted to produce national and association group estimates. For a detailed description of the weighting processes, see Abramson et al. (1996).

Standard Errors

Estimates found in the tables of this report are based on samples and are subject to sampling variability. Standard errors were estimated using a balanced repeated replications procedure that incorporates the design features of the stratified, clustered sample. The standard errors provide indications of the accuracy of each estimate. If all possible samples of the same size were surveyed under the same conditions, an interval of 1.96 standard errors below to 1.96 standard errors above a particular statistic would include the universe value in approximately 95 percent of the cases. Note, however, that the standard errors do not take into account the effects of biases due to item nonre sponse, measurement error, data processing error, or other systematic error. Estimates with large standard errors (coefficient of variation greater than 30 percent) should be interpreted with caution.

Cautions Concerning Change Estimates

Care must be taken in comparing the results in this report with results presented for earlier waves of SASS (e.g., McLaughlin, ODonnell, and Ries 1995), because some of the measured change (e.g., an 8 percent increase in the number of students receiving Chapter 1 services) may not be attributable to a change in the education system. Some of the change may be due to changes in the sampling frame, to a questionnaire item wording, or other changes detailed in Abramson et al. (1996).

Definitons and Analysis

A. Typologies of Private Schools

Private schools can be broadly divided into those that are religiously oriented and those that are not (nonsectarian schools) (McMillen and Benson 1991). Both religiously oriented and nonsectarian schools can be further meaningfully differentiated. By far the largest category of religiously oriented schools are those affiliated with the Catholic Church, and Catholic schools can be divided into parochial (associated with a parish), diocesan (associated with a diocese), and private orders. Other religiously oriented schools can be differentiated as conservative Christian, others affiliated with national church organizations, and unaffiliated schools. Among nonsectarian schools, two special types can be differentiated from the majority: schools focusing on the provision of special education and schools with a special programmatic emphasis (the largest group of which are the Montessori schools).

B. Suppression of Small Sample Data

Some of the figures presented in this report are based on thousands of respondents, others on a few. All statistics based on fewer than 10 respondents have been suppressed both to protect the privacy of the respondents and to avoid presentation of unstable results. In addition, and also to eliminate unstable results, all statistics that are based on fewer than 30 respondents and whose standard errors are more than 20 percent as large as their values are suppressed. Thus, for example, a statistic with a value of 10, based on 25 respondents, was not printed in this report if its standard error is greater than 2 (i.e., greater than 20 percent of 10).

C. Statistical Significance

Statements of relations between measures (e.g., between type of pri vate school and a characteristic such as school size) are included in the text only if they are statistically significant at = .05. All such state ments are either implicit or explicit comparisons of statistics for one group of schools with another. A comparison is statistically significant, for the purpose of this report, if the difference between statistics for two groups is greater than 1.96 times the standard error of the difference, approximated as the square root of the sum of squares of the estimated standard errors of the statistics for each of the groups of schools. That is, the large scale approximation to Students t statistic is used, with a two-tailed alpha level of .05.

When multiple statistical tests of significance are made on differences between figures in a table, then it is much more likely than .05 that some difference will be "significant" by chance. The Bonferroni correction can be applied to adjust for the increased likelihood of finding some difference significant. This correction involves definition of a "family" of comparisons and division of the alpha level (for reporting significance) by the number of comparisons in that family. A family is defined here as comparisons involving a single column of a table in the report. The definitions of family size for comparisons made in this report are one (1) when comparing the overall public-private graphic profile, except: for grade levels served, k = 3; for elementary services, k = 4; for secondary requirements, k = 5; for ratings of problems, k = 7; for teacher influence, k = 4; for satisfaction, k = 4; and for certification, k = 2. A family size of k = 9 is used for comparisons of typology profiles, such as Catholic parochial, with overall private school averages. In the Tables Section highlights, K = 1 for private versus public comparisons, k = 9 for typology comparisons, and k = 18 for private school type comparisons.

D. Factor Analysis of Teacher and Principal Attitudinal Items

A series of attitudinal items were included in the principals and teachers questionnaires for SASS. These items were factor analyzed to identify a set of factors that would provide more reliable information about teaches, administrators, and their schools than individual items. These are generally the same factors that were used in the 199091 private schools report (McLaughlin, ODonnell, and Ries 1995); in some cases, however, the factors are based on fewer items when item alternatives were not presented in 199394. For teachers in 199394, these items were:

In order to enhance the meaningfulness of the factors, the items were all re ordered in the positive direction and scaled on a scale from 0 to 10, where 0 refers to the "worst" response (e.g., strong agreement with a negative item) and 10 to the "best" response (e.g., "not a problem"). In computing the factor scores to report in 199394, all items included in a factor were given equal weight, and naming them was guided by communicativeness concerns. The factors are:

E. Definitions of Graphic Profile Elements

The private school survey items that were used for the private school graphics profiles are the following. Although the wordings are the same as for public schools or districts, the item numbers are different. The item numbers are given for the private school questionnaire.

Appendix AAcknowledgments  Appendix C